Meta Make-A-Scene: AI for image generation!

Under the name Meta "Make-A-Scene", the former Facebook recently unveiled its new image-generating AI. It is inspired by the classic DALL-E model. However, Mark Zuckerberg has set out to give this technology a new innovative twist. 

Do you want to know more about international commerce and e-commerce, and also get all the tips and news that might give your company an edge? Subscribe to Cross-Border Magazine and get your digital copy now for free!

The inspiration for Meta and its new AI

Throughout its history, Meta has repeatedly drawn inspiration from emerging technologies in order to launch its own projects. However, it is always looking to go one step beyond what already exists. 

In this sense, while DALL-E is the most popular artificial intelligence today, perfect for anyone looking to create artistic works by inserting certain words, Meta's researchers have decided to create a tool that could lead to more immersive worlds in the metaverse, making it possible for everyone, even those without the capacity for artistic creation, to generate their own works. This is how Make-A-Scene came about. 

Importantly, unlike DALL-E2, the new Meta Make-A-Scene AI does not aim for photorealism. According to the company, its images are already intended to have a more artistic feel, as if they were hand-painted. In this way, they argue, creators will be able to differentiate themselves from their competition. 

According to the Internet giant, what Meta is looking for with this development is that anyone can become an artist in the digital realm. They argue that this is a way of democratizing art in a way that had not been seen before, making it accessible even to those who do not have the talent to handle artistic techniques but wish to enter the world of art. 

The origin of Meta Make-A-Scene

One of the most interesting differences between Make-A-Scene and the traditional DALL-E is that it allows scenes to be shaped through the use of a sketch. It provides not only the textual input that DALL-E already offered, but also a window in which the user can specify the location and size of the elements of each composition, in a simple and easy to understand way. The sketch is used as a guide to the final result.

In this sense, Make-A-Scene allows you to capture the design of the scene, creating nuanced sketches as input. In addition, it helps to generate the design itself using only textual indications if the creator prefers. 

With this in mind, it can be said that it is a model that is particularly focused on learning the key aspects of images important to creators, such as animals or objects. Make-A-Scene also stands out for offering images with a resolution of 2,048 x 2,048 pixels. 

This AI is currently in closed testing, which means that it can only be used by users who have been invited by Meta to test it. Only these users will be able to use this technology until there is an official release or a public test.